Stavenga et al. Zoological Letters (2015) 1:14 DOI 10.1186/s40851-015-0015-2

RESEARCH ARTICLE Open Access

Combined pigmentary and structural effects tunewing scale coloration to color vision in theswallowtail butterfly Papilio xuthusDoekele G Stavenga1*, Atsuko Matsushita2 and Kentaro Arikawa2

Abstract

Butterflies have well-developed color vision, presumably optimally tuned to the detection of conspecifics by theirwing coloration. Here we investigated the pigmentary and structural basis of the wing colors in the Japanese yellowswallowtail butterfly, Papilio xuthus, applying spectrophotometry, scatterometry, light and electron microscopy, andoptical modeling. The about flat lower lamina of the wing scales plays a crucial role in wing coloration. In thecream, orange and black scales, the lower lamina is a thin film with thickness characteristically depending on thescale type. The thin film acts as an interference reflector, causing a structural color that is spectrally filtered bythe scale’s pigment. In the cream and orange scales, papiliochrome pigment is concentrated in the ridges andcrossribs of the elaborate upper lamina. In the black scales the upper lamina contains melanin. The blue scalesare unpigmented and their structure differs strongly from those of the pigmented scales. The distinct blue coloris created by the combination of an optical multilayer in the lower lamina and a fine-structured upper lamina.The structural and pigmentary scale properties are spectrally closely related, suggesting that they are undergenetic control of the same key enzymes. The wing reflectance spectra resulting from the tapestry of scales arewell discriminable by the Papilio color vision system.

Keywords: Papiliochrome, Melanin, Structural coloration, Interference reflector, Optical multilayer

IntroductionButterflies are highly visual animals, with well-developedcolor vision systems. The butterfly species whose visualcapacities have so far been investigated into the most detailis the Japanese yellow swallowtail butterfly, Papilio xuthus.By combining anatomy and intracellular electrophysiologywith spectrophotometry and related optical methods, theset of retinal photoreceptors has been shown to consist ofat least six different spectral classes, sensitive in the ultra-violet (UV), violet, blue, green, red and broad-band wave-length regions, respectively [1,2]. Additional behavioralinvestigations have revealed the butterfly’s extreme wave-length discrimination capacities. P. xuthus color vision ap-pears to be tetrachromatic, based on a set of UV, blue,green and red receptors [3].

* Correspondence: d.g.stavenga@rug.nl1Computational Physics, Zernike Institute for Advanced Materials, Universityof Groningen, Groningen NL9747AG, The NetherlandsFull list of author information is available at the end of the article

© 2015 Stavenga et al.; licensee BioMed CentrCommons Attribution License (http://creativecreproduction in any medium, provided the orDedication waiver (http://creativecommons.orunless otherwise stated.

Butterflies use their color vision to search for flower nec-tar and presumably for detecting and recognizing conspe-cifics [2]. We therefore considered that the wing colorationof P. xuthus might be tuned to its color vision system. Thisinitiated an optical analysis of the wing scales that coverthe wing substrate, so together forming a tapestry creatinga colorful and flashing signal displayed during flight [4].We recall here that the basic bauplan of a butterfly wingscale consists of two laminae. The lower lamina is a moreor less flat plate, connected by trabeculae to the elaboratearray of parallel ridges of the upper lamina, which are con-nected by crossribs. Generally the latter form open win-dows to the scale lumen, but when a membrane connectsthe crossribs, the windows are closed [5,6].The vast majority of butterfly wing scales contains pig-

ments, with melanin being most universal, causing brownand often very black scales [4]. The characteristic color pig-ments of papilionid butterflies belong to the class of papi-liochromes [7]. For instance, the pigment of the creamscales of P. xuthus, which absorbs exclusively in the

al. This is an Open Access article distributed under the terms of the Creativeommons.org/licenses/by/4.0) which permits unrestricted use, distribution, andiginal work is properly credited. The Creative Commons Public Domaing/publicdomain/zero/1.0/) applies to the data made available in this article,

Stavenga et al. Zoological Letters (2015) 1:14 Page 2 of 10

ultraviolet to violet wavelength region, has been charac-terized as papiliochrome II, and the pigment of the or-ange scales, which absorbs in a broader wavelengthrange, is presumably a related papiliochrome pigment[8]. Interestingly, females of the Eastern tiger swallowtail(Papilio glaucus) of North America can be either yellow(wild type) or black (melanic) [9,10]. The melanic formis a Batesian mimic of the distasteful Pipevine swallow-tail (Battus philenor), which is also black in overall color.In melanic females melanin replaces the backgroundyellow. The key enzyme involved is N-b-alanyl-dopamine-synthase (BAS), which shunts dopamine from the melaninpathway into the production of the cream-yellow coloringpapiliochrome. This enzyme is suppressed in melanicfemales, leading to abnormal melanization of a formerlyyellow area, and wing scale maturation is also delayed inthe same area. The changes in expression of a singlegene product thus result in multiple wing color pheno-types [9,10].It is commonly assumed that the color of pigmented

butterfly scales is caused by the irregular structures ofthe upper lamina acting as random scatterers and thatthe contained pigment determines the scale’s color byselective spectral filtering of the scattered light. How-ever, notably in scales with open windows, a consider-able fraction of incident light will pass the ridges andcrossribs and thus reach the lower lamina. It will bethere reflected dependent on the lower lamina’s struc-tural properties. The reflected light flux will partly leavethe scale unhampered through the windows, but it willalso partly traverse the ridges and crossribs of the upperlamina, where it will be spectrally filtered by the pig-ments present.Scales that do not harbor pigment can still be very

colorful when they have regularly arranged, nano-sizedstructures. Optical interference, which enhances lightreflection in a restricted wavelength range and sup-presses the reflection in the complementary wavelengthrange, then creates distinct colors [11,12]. The iconicexamples with structural coloration are the Morphobutterflies, which have scales with ridges exquisitelystructured into optical multilayers [5,6,13,14]. Amongthe papilionids, the Cattlehearts (Parides species) havescales furnished with multilayers or gyroid photoniccrystals [8,13,15].In the last decades many studies have been devoted to

unraveling the complex optics of the photonic struc-tures of butterflies, but it was only very recently recog-nized that in scales with a single-layered lower laminaoptical interference may also play a crucial role in scalecoloration. Research on the wing scales of B. philenorand a number of nymphaline butterfly species indicatedthat the lower lamina acts as an optical thin film,reflecting predominantly light in a limited wavelength

range [16,17]. In other words, the reflectance spectrumof a butterfly wing scale can be principally determinedby the pigmentation, alternatively by wave-optical ef-fects, or by a combination of the pigmentary and struc-tural scale properties [16-18].In the common butterfly wing scale the lower lam-

ina is a thin film. Because a thin film’s thickness pri-marily determines its reflectance spectrum, the directinference of the previous studies was that the differ-ently colored wing scales of butterflies may havelower laminae with different thicknesses. If so, thethickness of the lower lamina must then be undergenetic control and hence tunable and subject to evo-lutionary pressure. To substantiate the role of thelower lamina in butterfly wing scale coloration and toinvestigate the possible tuning of scale color andcolor vision, we chose P. xuthus for a detailed investi-gation of the structural and optical characteristics ofthe variously colored wing scales. To be able to dir-ectly assess the lower lamina’s thickness we per-formed anatomy, and this indeed yielded substantiallydifferent thickness values for the various scales. Tounderstand the scales’ different coloring methods intoquantitative detail, we applied spectrophotometry andscatterometry on single wing scales and performedoptical modeling. The obtained results reinforced ourview that pigmentation and scale structure are intim-ately connected and furthermore caused the hypothesisthat the wing colors are tuned to the swallowtail’s colorvision system.

Materials and methodsAnimalsSpecimens of the Japanese yellow swallowtail butterfly,Papilio xuthus, were from a laboratory culture derivedfrom eggs laid by females caught in Kanagawa, Japan.Photographs of a mounted specimen were taken with aNikon D70 digital camera equipped with an F Micro-Nikkor (60 mm, f2.8; Nikon, Tokyo, Japan) macroobjective.

Scale photographyPhotographs of single scales, obtained by gently pressingthe wings onto a microscope slide, in immersion fluidwere made with a Zeiss Universal Microscope, using epi-illumination through a Zeiss Epiplan 16x/0.35 objective(Zeiss, Oberkochen, Germany). Photographs were alsotaken of both sides, abwing and adwing, of single scalesglued to a glass micropipette. The adwing (or under) sidefaces the wing substrate in situ, and the abwing (or upper)side is the exposed side that is seen when observing thescales on the wing.

Stavenga et al. Zoological Letters (2015) 1:14 Page 3 of 10

AnatomyFor light and electron microscopy of the scales’ anat-omy, wing parts were prefixed in 2% paraformaldehydeand 2.5% glutaraldehyde in 0.1 mol l−1 sodium cacody-late buffer (CB, pH 7.3) or sodium phosphate buffer(PB, pH 7.3) for ~45 min at 20–25°C. Embedding inSpurr followed dehydration with a graded series ofethanol and infiltration with propylene oxide. Forlight microscopy, the tissues were cut into ~1 μmsections and observed with a BX51 (Olympus) micro-scope, equipped with a DP71 (Olympus) camera sys-tem. For transmission electron microscopy, the tissueswere cut into 50 nm sections, double-stained with ur-anyl acetate and lead nitrate, and observed with anH7650 transmission electron microscope (Hitachi,Tokyo, Japan), equipped with a CCD camera (Morada,Seika Digital Image Corp., Tokyo, Japan). For scan-ning electron microscopy, the wing pieces were sput-tered with platinum and observed with a JSM-6490LV(JEOL, Tokyo, Japan).

SpectroscopyReflectance spectra of single wing scales in situ or fromisolated scales taken from different wing areas andattached to a glass micropipette were measured with amicrospectrophotometer (MSP), being a Leitz Ortholuxmicroscope (Leitz, Wetzlar, Germany) connected to anAvaSpec 2048–2 CCD detector array spectrometer(Avantes, Eerbeek, Netherlands), with light supplied by axenon arc light source. The microscope objective was anOlympus LUCPlanFL N 20x/0.45 (Olympus, Tokyo,Japan). A white diffuse reflectance tile (Avantes WS-2)was used as a reference. Absorbance spectra of singlescales, immersed in immersion oil or water were alsomeasured with the MSP. Reflectance measurements ofvarious wing areas of intact butterfly wings were mea-sured with a bifurcated probe (Avantes FCR-7UV200;Avantes, Eerbeek, Netherlands); the light source was adeuterium-halogen lamp (AvaLight-D(H)-S), and thewhite diffuse reflectance tile served as reference. It is im-portant to note that the diffuse reference yields for specu-lar objects reflectance spectra where the shape isrepresentative but the amplitude is overestimated.

Imaging scatterometryThe spatial reflection characteristics of the wing scaleswere studied with an imaging scatterometer. An isolated,single scale attached to a glass micropipette was positionedat the first focal point of the ellipsoidal mirror of the im-aging scatterometer [19]. The scatterograms were obtainedby focusing a white light beam with narrow aperture (<5°)onto at a small circular area (diameter ~13 μm) of thescale, and the spatial distribution of the far-field scatteredlight was then monitored. A flake of magnesium oxide

served as a white diffuse reference object (for further de-tails, see [19]).

Modeling of reflectance spectraThe reflectance spectrum of the lower lamina of the pig-mented scales was calculated for normally incident lightusing the classical Airy formulae for thin films [18,20], as-suming that the scale’s lower lamina acts as an opticalthin film with thickness depending on the scale type.The reflectance spectrum of the lower lamina of bluescales was modeled by treating it as an optical multilayer,using a matrix-transfer formalism [18].

ResultsWing colors and pigmentationThe wing colors of P. xuthus can be distinguished intothe following main categories: yellow, orange, black,and blue (numbered 1–4 in Figure 1A). These colorsare observable on both the dorsal (Figure 1A) and theventral wings (Figure 1B). A rare, fifth scale categorywith a yellow-ochre color exists at the ventral hindw-ings (#5, Figure 1B). Because the colors originate fromthe lattice of wing scales, as a first characterization wemeasured reflectance spectra from individual scales inthe differently colored areas (Figure 1C).The spectra suggested the presence of various pig-

ments, and to investigate this more closely we pickedup scales from the wing and measured the absorbancespectra of single scales with a microspectrophotometer(Figure 1D). To suppress possible structural colora-tions, we immersed the scales in immersion oil. Thesingle scales obtained from the dorsal yellow wing areas(Figure 1A, #1) had an unsaturated yellow, cream color.They prominently absorbed only in the ultraviolet toviolet wavelength range, with peak absorbance of ~0.6(c in Figure 1D). The responsible pigment has beencharacterized as papiliochrome II [7]. Scales of the ven-tral yellow wing areas (Figure 1B) had a much palercream color. Their absorbance spectrum had the sameshape as that of the dorsal cream scales, but the peakabsorbance was no more than ~0.2 (p in Figure 1D),meaning that the density of the papiliochrome II pig-ment in the ventral scales was distinctly lower thanthat in the dorsal scales. The scales obtained from theorange wing areas absorbed well into the visible wave-length range and showed considerable peak absor-bances, approaching 1 (Figure 1A, #2; o in Figure 1D).The absorbance of single black scales was substantialthroughout a very broad wavelength range, clearly due tohigh concentrations of melanin pigment (Figure 1A, #3; kin Figure 1D). Very differently, the blue scales had anegligible absorbance (Figure 1A, #4; b in Figure 1D),meaning that the blue color must have a structuralorigin.

Figure 1 Wing coloration of Papilio xuthus. A, B Dorsal and ventral view of a spring form female. The numbers refer to the wing areas from wherescales were taken (scale bar: 2 cm). C Reflectance spectra measured with a microspectrophotometer from single scales in situ of the numbered areas ofA and B. D Absorbance spectra of scales embedded in immersion oil; c: a cream scale (from area 1); o: orange scale (area 2); k: black scale (area 3), b:blue scale (area 4); scales of area 5 are not included (scale bar: 20 μm). The thin curves are from small measurement areas (15 × 15 μm2) near the tipand middle parts of a few different scales. The bold curves represent the averaged spectra (of 4, except for the black: 2). The deep-yellow curves arefrom cream-colored (c) scales on the dorsal hindwing, from area 1, and the yellow spectra are from pale-cream-colored (p) scales located on the ventralhindwing, opposite to the dorsal area 1.

Stavenga et al. Zoological Letters (2015) 1:14 Page 4 of 10

Structure of the wing scales of P. xuthusWhereas the color of the blue scales must be fully struc-tural, the color of the pigmented scales might not befully determined by the pigment and some structuralcoloration could be present also. To clarify the possiblecontribution of the scales’ structures to their color, weinvestigated the anatomy of the scales. As can be directlyobserved with a binocular microscope, the scales existon both sides of the wing and partly overlap each other,and this causes the partly stacked scales in the lightmicroscopical cross-sections of the various wing partscarrying the four main scale types (Figure 2A-D). Thelight micrographs showed that the scales of P. xuthusconsist of a more or less flat lower lamina, which facesthe wing substrate, and a highly structured upper lam-ina. The lower and upper lamina are therefore alsocalled the adwing and abwing lamina, respectively (latinad: to, on; ab: away from; see inset Figure 3C, columnIII). The cream, orange and black scales appeared tohave their dense pigmentation concentrated in the

upper lamina, which made them easily recognizable inunstained sections (Figure 2A-C). The pigment con-centration in the blue scales was too low for recogni-tion of the scales, however, and therefore the section ofFigure 2D, showing a blue scale with two underlyingblack scales, was stained with Azur II.Transmission electron micrographs of the pigmented

scale types yielded very similar structures (Figure 2E-G).Quite strikingly, the thickness of the lower lamina var-ied among these scales, with notably the orange scaleshaving a very thin lower lamina (Figure 2F). The elec-tron micrographs of all pigmented scales showed muchless contrastful lower laminae than the elaboratelystructured upper laminae, in agreement with the pre-dominant or even exclusive presence of pigment in theupper laminae (Figure 2E-G). Surprisingly, the bluescales showed a very aberrant anatomy. Both the lowerand upper lamina appeared to have a somewhat similarholey structure, with the lower lamina essentially con-sisting of two membranes, which were only locally

Figure 2 Anatomy of cream (c), orange (o), black (k), and blue (b) wing scales. A-D Light microscopy. E-H Transmission electron microscopy.I-L Scanning electron microscopy. In A and C the wing membrane substrate (arrows) has on both sides cream and black scales, respectively.The section of D was stained with Azur II to visualize the unpigmented blue scale, which is situated above two black scales. The upper lamina has in E(asterisk) and G a high electron density, but in the lower lamina (E, white arrow) the electron density is low. The lower lamina of the blue scales consistsof two membranes (H, arrows) with local connections. Membranous structures can be seen between crossribs (e.g., arrowhead in J). Scale bars: A-D20 μm. E-L 2 μm.

Stavenga et al. Zoological Letters (2015) 1:14 Page 5 of 10

connected (Figure 2H). The lower lamina of the bluescale thus is not a thin film but rather a multilayer.Scanning electron microscopy of the upper lamina of

the various scales showed in all cases a structure charac-teristic for papilionid butterflies (Figure 2I-L). The par-allel ridges, with interdistance 1–2 μm, were connectedwith irregularly arranged crossribs, with often closedwindows, that is, connected by a membrane. Taken to-gether, the anatomical studies showed that pigment, ifpresent, is at least for the major part expressed in theupper lamina, that the lower lamina of the pigmentedscales is basically a thin film and thus will act as aninterference reflector, and that structural effects deter-mine the color of the blue scales.

Spatial and spectral characteristics of the scale reflectionsTo further investigate the contribution of the scalestructures to the scale coloration, we applied imagingscatterometry and microspectrophotometry on both theabwing (upper) side and adwing (under) side of singlescales. The photographs and scatterograms of the inves-tigated scales (columns I and II of Figure 3) indicatedthat the optical properties of the two opposite sides of

the scale rather differ. The abwing side was alwaysmatte. Accordingly, scatterometry using narrow-beamillumination produced wide-angled, diffuse scattero-grams; in most cases also a narrow, colored line ap-peared, revealing diffraction on the array of parallelridges (Figure 3B-D, column II, ab). The adwing side ofthe scales was generally rather glossy, indicating thatthe lower lamina acted mirrorlike, and the correspond-ing scatterograms accordingly showed a spatially re-stricted pattern. If the lower laminae had been perfectthin films, the scatterograms would have shown singledots, but the patterns were in fact slightly blurred. In-deed, the photographs of the adwing side (Figure 3A-D,column I, ad) indicated that the lower lamina of thescales was not perfectly flat but slightly ripply (see alsoFigure 2A-H).The different scale types had each very specific reflect-

ance spectra (Figure 3A-D, column III). The scattero-grams of the adwing side of the cream, orange and blackscales indicate that the lower lamina’s thin film reflec-tions must determine the adwing reflectance spectrum.On the other hand, the reflectance spectrum of theirabwing side may be expected to be greatly determined

Figure 3 (See legend on next page.)

Stavenga et al. Zoological Letters (2015) 1:14 Page 6 of 10

(See figure on previous page.)Figure 3 Photographs, scatterograms, and reflectance spectra of single scales, glued to a thin glass micropipette, with modeled spectra. A-D Acream (c), orange (o), black (k), and blue (b) scale, respectively. In each panel, photographs (column I) and scatterograms (column II) of the upperand lower side (abwing: ab, and adwing: ad; see inset C, column III) of the scales are shown above and below, respectively. The red circles in thescatterograms indicate scattering angles of 5°, 30°, 60°, and 90°. The reflectance spectra were measured with respect to a diffusely reflecting whitereference, and thus yielded too high values for directionally reflecting media. A-C, column IV, present reflectance spectra modeled with opticalthin film and multilayer theory for five thin films consisting of butterfly chitin in air with different thicknesses (dotted curves) and their average(dashed bold curve), for the cream (c), orange (o), and black (k) scales, respectively. The averaged spectra, when multiplied with a spectrumrepresentative for the transmittance spectrum of the scale’s pigment (derived from Figure 1D) yielded the continuous bold curves. The thickness(in nm) of the five scales was: 210 + 10i (c), 115 + 5i (o), 170 + 10i (k), with i =1-5. D column IV, presents reflectance spectra of three chitinousthin films in air with thickness 190, 200, and 210 nm, and reflectance spectra of a stack of three parallel layers in air, with thicknesses 60, 80 and60 nm, where the upper and lower layer had the refractive index of chitin. The refractive index of the middle layer was taken to be a weightedaverage of the refractive indices of chitin and air with ratio 1:2 (200a) and ratio 1:1 (200b).

Stavenga et al. Zoological Letters (2015) 1:14 Page 7 of 10

by the absorbance spectra of the various pigments.Strikingly, however, the reflectance spectra of both sidesof the cream and orange scales were similar, indicatingthat thin film reflectance and pigment absorption arespectrally tuned. In the black scales the reflectance ofthe abwing side was very minor throughout the wholeultraviolet and visible wavelength range, clearly due tothe strongly absorbing melanin (Figure 1D). Neverthe-less, the adwing reflectance spectrum of the black scalesfeatured a distinct, blue-green peaking spectrum, verycharacteristic of an optical thin film. In fact, all adwingreflectance spectra of Figure 3 suggested the presence ofan interference reflector. To substantiate this we per-formed optical modeling, using the measured pigmentspectra and the anatomy of the four main scale types.

Modeling scale reflectance spectraThe transmission electron micrographs of the cream,orange and black scales showed that their lower laminawas a single layer with thickness between 100 and250 nm. Using classical thin film theory we calculatedreflectance spectra for the lower lamina of the differentscales (Figure 3A-C, column IV). We assumed the ab-sence of pigment in the lower lamina, as suggested byboth the light and transmission-electron micrographs,and we took for the refractive index of the lower lami-na’s material, chitin, the value determined for the glassscales of Graphium sarpedon [21]. The TEM micro-graphs (Figure 2E-G) showed that the scale thicknesseswere not constant, and therefore we calculated for eachscale type reflectance spectra for five values of thethickness, d; for the cream scale d = 220, 230, …260 nm(Figure 3A, column IV), for the orange scale d = 120,125 …140 nm (Figure 3B, column IV), and for the blackscale d = 180, 190,… 220 nm (Figure 3C, column IV).For each scale type we averaged the five reflectancespectra (Figure 3A-C, column IV, dotted curves), whichyielded for the cream scale a distinctly blue-green peakingspectrum (at ~500 nm), for the orange scale an increas-ing reflectance in the visible wavelength range, andfor the black scale a blue-peaking reflectance, with

maximum near 410 nm (Figure 3A-C, column IV,dashed curves). The averaged spectra resembled themeasured reflectance spectra of the adwing sides of thescales, but a close correspondence was not reached, ex-cept perhaps for the black scale. An obvious reason forthe discrepancy is that in the calculations only the thinfilm of the lower lamina was taken into account, whilethe other scale components may also contribute to theadwing reflectance spectra. Especially in the longerwavelength range, where pigment absorption is minor,background scattering from the upper lamina will raisethe reflectance.For the adwing reflectance spectrum of the blue scales,

the situation is even more complex. The transmissionelectron micrograph of Figure 2H showed that the lowerlamina is not a single thin film, but rather a multilayer,consisting of two thin membranes separated by an airgap, with pieces of membrane that connect the twoouter layers. The total thickness of the lower lamina wasabout 200 nm, with thickness of the two membranousouter layers ~60 nm and that of the middle layer, con-sisting mostly of air, ~80 nm. To calculate the reflect-ance spectrum of this structure, we assumed that thetwo membranes consisted of chitin and that the refract-ive index of the middle layer is the weighted average ofchitin and air. We considered two cases, with the mem-brane to air ratio of the middle layer being 1:2 and 1:1,respectively. Using a matrix-transfer procedure for opticalmultilayers we calculated that the reflectance spectrum ofthe three-layered lower lamina has a distinct peak near400 nm, with amplitude increasing with decreasing mem-brane content (Figure 3D, column IV; for comparison wehave included the reflectance spectra of single thin filmsconsisting of chitin with thicknesses 190, 200, and210 nm). Clearly, the splitting of a solid thin film intothree layers (chitin-air-chitin) causes, in addition to aslight shift in peak wavelength, a most substantial increasein reflectance (note the difference in ordinate in Figure 3Dcolumn IV with that of Figure 3A-C column IV).For the interpretation of the reflectance spectra mea-

sured from the abwing side of the various scale types,

Figure 4 Wing coloration and color discrimination of Papilio xuthus.A Reflectance spectra measured with a bifurcated probe from wingareas 1–5. B Wavelength discrimination as a function of wavelength,Δλ(λ), following from behavioral experiments (after Figures 1b and 4dof [3]) and the spectral sensitivity, S(λ), measured by intracellularrecordings, of the four photoreceptor types that were concluded toparticipate in spectral discrimination; the normalized S(λ)-values weremultiplied with a factor 10 to facilitate visualization. The dottedvertical lines, with interdistance 80 nm, are at the wavelengths whereΔλ was about minimal.

Stavenga et al. Zoological Letters (2015) 1:14 Page 8 of 10

the properties of the upper lamina have to be included.However, the structure of the upper lamina is too com-plex for a simple computational treatment. The maineffect of the highly convoluted upper lamina will be adiffuse, more or less wavelength-independent reflection,which is spectrally filtered by the expressed pigment. Inthe cream, orange and black scales, incident light thathas traversed the upper lamina and is reflected by thelower lamina will be spectrally filtered by the pigmentin the upper lamina. We have assessed the spectral fil-tering for each scale type by multiplying the averagedthin film reflectance spectra of Figure 3A-C (columnIV, dashed curves) with the transmittance spectrumfollowing from the absorbance spectrum of the scale’spigment (Figure 1D). The resulting spectra showed aslightly suppressed reflectance for the cream and or-ange scales and a substantially reduced reflectance forthe black scales (Figure 3A-C, column IV, continuouscurves). We have to emphasize that the latter spectraare only one component of the reflectance spectrummeasured from the abwing side. Another component isthe direct (spectrally filtered) backscattering by theridges and crossribs. Furthermore, when the windowsare closed, the membrane between the crossribs willlocally also act as a thin film reflector and thus add tothe reflectance in a wavelength-dependent manner.

Wing coloration and color visionWe have found that the wings of P. xuthus has scaleswith a restricted set of colors. Extensive research dem-onstrated that the eyes are furnished with a restricted setof spectral photoreceptors. We therefore conjecturedthat the wing colors are intimately related to the spectralsensitivities of the photoreceptors. To test this hypoth-esis, we have measured reflectance spectra of the variouscolored wing areas, using a bifurcated probe (Figure 4A),and compared the spectra with wavelength discrimin-ation (Δλ) data obtained from behavioral studies [3].The latter data showed three wavelengths where wave-length discrimination was minimal: ~420, 500 and580 nm (Figure 4B). The minima separate the visiblespectrum of P. xuthus into four wavelength regionswhere the four photoreceptors, assumed to form thebasis of the spectral discrimination system, are max-imally sensitive, i.e. the ultraviolet (UV), blue (B), green(G), and red (R) sensitive photoreceptors. ComparingFigure 4A and 4B suggests that the wing colors are welldiscriminable by the color vision system.

DiscussionWing scale colors are due to spectrally tuned pigmentfilters and interference reflectorsThe wings of the Japanese yellow swallowtail P. xuthusbutterfly are colored by scales that contain various

pigments and consist of nanostructured elements(Figures 1 and 2). Due to their wavelength-selective ab-sorption, pigments generally play an important role in col-oration. It hence has been commonly assumed that thecolor of the cream and orange scales is principally causedby their specific papiliochromes, especially because theabwing and adwing side of the scales are similarly colored(Figure 3A, B, column I), and indeed the reflectance spec-tra measured from both scale sides differ only slightly(Figure 3A, B, column III). However, the scatterogramsdemonstrate that the lower lamina acts as a thin filminterference reflector (Figure 3A, B, column II). The re-markable finding that the pigmentation of the upper lam-ina and the thin film optics of the lower lamina yield verysimilar reflectance spectra forces the conclusion that thetwo different optical mechanisms are spectrally tuned.

Stavenga et al. Zoological Letters (2015) 1:14 Page 9 of 10

Optimal spectral tuning can even be argued for the blackscales, which contain the broad-band absorbing melanin.Adwing illumination yielded a blue-peaking reflectancespectrum (Figure 3C, column III), which makes sense, be-cause with abwing illumination part of the incident lightwill reach the lower lamina. The resulting blue-peaking re-flectance will be effectively suppressed by the melanin inthe upper lamina, because melanin absorbs more stronglyat shorter than at longer wavelengths (Figure 1D). Hence ablacker scale is realized when the lower lamina reflectsmaximally in the blue rather than at longer wavelengths(Figure 3C, column III).The reflectance spectrum of the blue scales shows a

shallow valley in the long-wavelength range, similarly asthe rare yellow-ochre scales (Figure 1C #4, 5). Presumablythe ultrastructure of the blue and yellow-ochre scales issimilar, but the low reflectance around 400 nm of the lat-ter scales indicates that they contain papiliochrome II.

Genetic control of scale pigment and structureThe expression of pigments in the wing scales has beenstudied in a few related papilionids. In the CommonMormon swallowtail butterfly, Papilio polytes, whose fe-males mimic the unpalatable Common Rose swallowtailbutterfly, Pachliopta aristochiae, the pale-yellow wing re-gions in non-mimetic females consist of kynurenine andN-b-alanyldopamine (NBAD). The kynurenine/NBADbiosynthetic genes are absent in mimitic females [22]. Fur-thermore, Nishikawa et al. [22] found that kynurenine/NBAD synthesis and Toll signaling genes were upregulatedin the red spots specific to mimetic female wings. In theEastern tiger swallowtail, Papilio glaucus, the wild type fe-males are yellow, but the black, melanic form is a Batesianmimic of the distasteful Pipevine swallowtail (B. philenor),which has an overall black color [9,10]. In melanic females,N-b-alanyl-dopamine-synthase (BAS) is suppressed, so thatmelanin replaces papiliochrome II.As in the present study on P. xuthus scales, in our previ-

ous study on B. philenor wing scales [16] we found a clearcorrelation of the ad- and abwing reflectance spectra ofthe various colored wing scales, reinforcing the hypothesisthat both scale structure (and thus lower lamina thickness)and pigment expression are under genetic control of thesame key enzymes. Similar conclusions were drawn forthe wing scales of a few nymphaline butterfly species, asthere the absorbance spectra of the pigment in the upperlamina and the reflectance spectra of the lower lamina ap-peared to be tuned also [17]. The latter spectra could bewell described by thin film modeling, and therefore thelower lamina of the scales of the nymphalines were as-sumed to be a single layer. However, detailed transmissionelectron microscopy was not performed in the earlierstudies. The present anatomy confirmed that the lowerlaminae of the pigmented scales of P. xuthus were single

layers, but surprisingly the lower lamina of the blue scalesdeviated from a single thin film and appeared to be a multi-layer. The anatomy of the scales of other butterfly speciestherefore deserves to be examined more closely.An architecture very similar to that of the blue scales of

P. xuthus, but even more elaborate, exists in the blue scalesof lycaenids, the blues [23-26]. In these scales, the lumen isfilled with several evenly spaced membranes, connected byminor struts, that together create a brightly reflectingmultilayer. Another method of creating bright blue reflec-tors is the extensive folding into multilayers of the ridgesfound in Morpho butterflies [5,6,13,24].

The wing scale colors are tuned to the butterfly’s colorvisionThe coloration of the wing scales of P. xuthus appears tobe optimized in different ways. The unpigmented bluescales have an increased blue-peaking reflectance due to anincreased number of layers in the lower lamina. Also theupper lamina is fine-structured (Figure 2H). The variouspapiliochromes are expressed in scales with a lower laminathat reflects maximally in the wavelength range where pig-ment absorption is minimal, suggesting that the structuralproperties of the lower lamina and the pigmentation areunder the same genetic control. Very recently, Monteiroand colleagues demonstrated in the satyrine butterfly Bicy-clus anynana that the thickness of the lower lamina of cer-tain scales can be specifically modified in a few generationsby selectively breeding butterflies with the scale color as se-lection criterion [27]. Presumably therefore, the restrictedset of wing colors of P. xuthus is the result of evolutionaryselection processes.The obvious question, what drives the tuning of the scale

colors, suggests an obvious answer, namely selection by in-traspecific visual recognition. Figure 4 indicates that theyellow wing areas excite the UV receptor to a minor extentand the B, G and R receptors strongly, while the orangewing parts negligibly stimulate the UV and B receptors andmuch more the G and R receptors; the blue wing spots dothe opposite. P. xuthus lacks red scales, present in manyother papilionid butterflies, which presumably contain an-other papiliochrome pigment [8,22]. If papilionid butter-flies have similar color vision capacities, the red wing areaswill stimulate virtually exclusively the red receptors.Butterfly wings are covered by a tapestry of overlapping

scales and therefore the color of the various wing areas isthe result of a summation of reflections on the stacks ofwing scales on both sides of the wing together with thewing substrate proper (Figure 2A-C). Consequently, the re-flectance spectra are much higher than those obtainedfrom the individual scales. For instance, the stacked creamscales create a yellowish color, which is more saturatedthan that of single scales (Figures 3 and 4). This does nothold for the blue scales, however, because they are stacked

Stavenga et al. Zoological Letters (2015) 1:14 Page 10 of 10

upon black scales (Figure 2D) as otherwise the blue struc-tural color would become desaturated [16,17].When mating, male butterflies find females due to the yel-

low and black areas, which create a characteristic bandingpattern with high color as well as brightness contrast [28].Interestingly, the most colorful patches are on the lower re-gion of the hindwings on both wing sides (Figure 1A,B).This region is not covered when sitting, neither with opennor with closed wings. Even when the butterflies forage onflowers, the hindwings do not move much compared to theforewings. As the color patterns of both forewings andhindwings are generally well displayed they will thus servefor conspecifics as a conspicuous recognition signal, allow-ing optimal detection, thanks to well-tuned scale colors.

Competing interestsThe authors declare that they have no competing interests.

Authors’ contributionsDGS and KA designed the study. DGS performed the optical experimentsand modeling, and drafted the manuscript. AM carried out the anatomicalstudies (light and electron microscopy). All authors read and approved thefinal manuscript.

AcknowledgementsThis study was financially supported by AFOSR/EOARD (grant FA8655-08-1-3012to D.G.S.) and JSPS (grant-in-aid for scientific research A 26251036 to K.A.). BodoWilts and two anonymous referees provided constructive criticisms.

Author details1Computational Physics, Zernike Institute for Advanced Materials, Universityof Groningen, Groningen NL9747AG, The Netherlands. 2Laboratory ofNeuroethology, SOKENDAI (The Graduate University for Advanced Studies),Hayama 240-0115, Japan.

Received: 3 February 2015 Accepted: 2 April 2015

References1. Arikawa K. Spectral organization of the eye of a butterfly, Papilio. J Comp

Physiol A. 2003;189:791–800.2. Kinoshita M, Arikawa K. Color and polarization vision in foraging Papilio.

J Comp Physiol A. 2014;200:513–26.3. Koshitaka H, Kinoshita M, Vorobyev M, Arikawa K. Tetrachromacy in a

butterfly that has eight varieties of spectral receptors. Proc Roy Soc B.2008;275:947–54.

4. Nijhout HF. The Development and Evolution of Butterfly Wing Patterns.Washington: Smithsonian Institution Press; 1991.

5. Ghiradella H. Hairs, bristles, and scales. In: Locke M, editor. Microscopic anatomyof invertebrates, Vol 11A: Insecta. New York: Wiley-Liss; 1998. p. 257–87.

6. Ghiradella H. Insect cuticular surface modifications: scales and otherstructural formations. Adv Insect Physiol. 2010;38:135–80.

7. Umebachi Y. Papiliochrome, a new pigment group of butterfly. Zool Sci.1985;2:163–74.

8. Wilts BD, IJbema N, Stavenga DG. Pigmentary and photonic colorationmechanisms reveal taxonomic relationships of the Cattlehearts(Lepidoptera: Papilionidae: Parides). BMC Evol Biol. 2014;14:160.

9. Koch PB, Keys DN, Rocheleau T, Aronstein K, Blackburn M, Carroll SB, et al.Regulation of dopa decarboxylase expression during colour pattern formationin wild-type and melanic tiger swallowtail butterflies. Development.1998;125:2303–13.

10. Koch PB, Behnecke B, ffrench-Constant RH. The molecular basis of melanismand mimicry in a swallowtail butterfly. Curr Biol. 2000;10:591–4.

11. Kinoshita S. Structural colors in the realm of nature. Singapore: WorldScientific; 2008.

12. Biró LP, Kertész K, Vertésy Z, Márk GI, Bálint Z, Lousse V, et al. Livingphotonic crystals: Butterfly scales - Nanostructure and optical properties.Mat Sci Eng C. 2007;27:941–6.

13. Vukusic P, Sambles JR. Photonic structures in biology. Nature. 2003;424:852–5.14. Kinoshita S, Yoshioka S, Fujii Y, Osanai M. Photophysics of structural color in

the Morpho butterflies. Forma. 2002;17:103–21.15. Michielsen K, Stavenga DG. Gyroid cuticular structures in butterfly wing

scales: biological photonic crystals. J R Soc Interface. 2008;5:85–94.16. Stavenga DG, Leertouwer HL, Wilts BD. The colouration toolkit of the

Pipevine Swallowtail butterfly, Battus philenor: thin films, papiliochromes,and melanin. J Comp Physiol A. 2014;200:547–61.

17. Stavenga DG, Leertouwer HL, Wilts BD. Colouration principles of nymphalinebutterflies - thin films, melanin, ommochromes and wing scale stacking.J Exp Biol. 2014;217:2171–80.

18. Stavenga DG. Thin film and multilayer optics cause structural colors ofmany insects and birds. Mat Today Proc. 2014;1:109–21.

19. Stavenga DG, Leertouwer HL, Pirih P, Wehling MF. Imaging scatterometryof butterfly wing scales. Opt Express. 2009;17:193–202.

20. Yeh P. Optical waves in layered media. Hoboken NJ: Wiley-Interscience; 2005.21. Leertouwer HL, Wilts BD, Stavenga DG. Refractive index and dispersion of

butterfly scale chitin and bird feather keratin measured by interferencemicroscopy. Opt Express. 2011;19:24061–6.

22. Nishikawa H, Iga M, Yamaguchi J, Saito K, Kataoka H, Suzuki Y, et al.Molecular basis of wing coloration in a Batesian mimic butterfly, Papiliopolytes. Sci Reports. 2013;3:3184.

23. Schmidt K, Paulus H. Die Feinstruktur der Flügelschuppen einigerLycaeniden (Insecta, Lepidoptera). Z Morph Tiere. 1970;66:224–41.

24. Ghiradella H. Light and color on the wing: structural colors in butterfliesand moths. Appl Optics. 1991;30:3492–500.

25. Bálint Z, Kertész K, Piszter G, Vertésy Z, Biró LP. The well-tuned blues: therole of structural colours as optical signals in the species recognition of alocal butterfly fauna (Lepidoptera: Lycaenidae: Polyommatinae). J R SocInterface. 2012;9:1745–56.

26. Ingram AL, Parker AR. A review of the diversity and evolution of photonicstructures in butterflies, incorporating the work of John Huxley (The NaturalHistory Museum, London from 1961 to 1990). Phil Trans R Soc B.2008;363:2465–80.

27. Wasik BR, Liew SF, Lilien DA, Dinwiddie AJ, Noh H, Cao H, et al. Artificialselection for structural color on butterfly wings and comparison withnatural evolution. Proc Natl Acad Sci U S A. 2014;111:12109–14.

28. Hidaka T, Yamashita K. Wing color pattern as the releaser of matingbehavior in the swallowtail butterfly, Papilio xuthus L. (Lepidoptera:Papilionidae). Appl Ent Zool. 1975;10:263–7.

Submit your next manuscript to BioMed Centraland take full advantage of:

• Convenient online submission

• Thorough peer review

• No space constraints or color figure charges

• Immediate publication on acceptance

• Inclusion in PubMed, CAS, Scopus and Google Scholar

• Research which is freely available for redistribution

Submit your manuscript at www.biomedcentral.com/submit